sign in sign up
<<
Home , Agricultural expansion, Mutualism (biology)

2713 No. of Pages 10

Trends in &

Opinion Deciphering the –Production Mutualism in the Global Debate

Ralf Seppelt ,1,2,* Channing Arndt,3 Michael Beckmann,1 Emily A. Martin,4 and Thomas W. Hertel5

Without changes in consumption, along with sharp reductions in food waste and Highlights postharvest losses, agricultural production must grow to meet future food de- Increasing demands for agricultural mands. The variety of concepts and policies relating to yield increases fail to inte- commodities are resulting in more in- grate an important constituent of production and nutrition – biodiversity. tensely managed landscapes. This is at odds with biodiversity conservation and We develop an analytical framework to unpack this biodiversity-production mutu- largely ignores farmland biodiversity’s alism (BPM), which bridges the research fields of ecology and agroeconomics and supporting function for high and stable makes the trade-off between food security and protection of biodiversity explicit. yields. By applying the framework, the incorporation of agroecological principles in fi An overhaul of agroeconomic models to global food systems are quanti able, informed assessments of green total factor account for the biodiversity-production (TFP) are supported, and possible lock-ins of the global food system mutualism is urgently needed to answer through overintensification and associated can be avoided. a question of utmost importance: How do we manage the resources of our planet in such a way that we produce Consequences of Increasing Food Production enough healthy food without destroying our life-support system? The quest for greater crop output for food and non-food products [1–3] leads to both an increase in use and an increase in yields, typically achieved through an intensification of culti- A comprehensive analytical framework is vation methods that help to close yield gaps (see Glossary)[4,5,62]. This, in turn, leads to a loss of provided that accounts for multitrophic biodiversity-production processes; brid- biodiversity in agricultural landscapes [7] and increases pressure on natural diversity [8], which con- ges disciplinary boundaries between tinues declining despite ongoing efforts for protection [9]. The avoidance of food waste and dietary , , economics, changes offer two demand-side options to reduce pressure on food production [10], but these have and conservation science; and eluci- thus far not been achieved at the macro level [11]. dates the strong interactions of ecosys- tem functioning with food security and malnutrition. Biodiversity is a crucial component of functions that are essential for agricultural production, such as soil fertility, , and biocontrol (i.e., the control of pests by their nat- 1 – – UFZ Helmholtz Centre for ural enemies) [12 15]. This interdependence of biodiversity and agricultural production has led to a Environmental Research, Department of variety of concepts that aim to optimize the management of agricultural landscapes, balancing yields, Computational , biodiversity, and sustainability [16]. These concepts make use of agroecological principles [17], Leipzig, Germany 2Institute of Geoscience and promote [18] suggest ecological intensification [19], or sustainable intensifi- Geography, Martin-Luther-University cation (SI) [20], compare land sharing and land sparing concepts [21–23] and take the perspec- Halle-Wittenberg, Halle (Saale), Germany 3 tive of managing a coupled socioecological system (SES). While there is much published research on International Food Policy Research Institute, Washington, DC 20005, USA the SES concept, quantitative, empirically-based, and model-based implementations are largely 4Zoological Biodiversity, Institute of lacking or their improvement through process-based validation is pending [24]. As a result, it is cur- Geobotany, Leibniz University Hannover, rently not possible to capture and quantify the biodiversity-production mutualism (BPM) in its entirety. Hannover, Germany 5Department of Agricultural Economics, However, recent research provides the necessary basis for such a comprehensive, quantitative, Purdue University, West Lafayette, IN process-based understanding of the BPM concept [15,24,25]. 47907, USA

A Multidisciplinary Perspective on the Relationship between and Biodiversity How is Biodiversity Affected by Cropland Management? Management of agricultural landscapes serves the provisioning of agricultural goods and has had *Correspondence: mostly negative impacts on biodiversity [7,8]. This occurs through both the expansion and [email protected] (R. Seppelt).

Trends in Ecology & Evolution, Month 2020, Vol. xx, No. xx https://doi.org/10.1016/j.tree.2020.06.012 1 © 2020 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Trends in Ecology & Evolution

intensification of cultivated areas, which in turn warps the composition of landscapes and their struc- Glossary fi ture. Conventional intensi cation to increase yields is typically done by homogenizing the landscape Agroecological principles: are (fewer, larger fields), increasing inputs (labor, , irrigation, chemicals), and/or augmenting har- supposed to contribute to transforming vest intensities [26–29]. Conventional intensification typically leads to a loss of present and food systems by applying ecological fi a change in the composition of communities, for example, in the form of a general decrease in abun- principles (ecological intensi cation) to ensure a regenerative use of natural dance [7,8,18]. There is evidence of a long-term decline in species due to loss and ag- resources while addressing the need for ricultural intensification [30,31]. This can also be associated with a proportionally greater socially equitable food systems. While of pest species due to a reduction in biological [32]. Ecological intensification directly or the focus initially was on understanding field-level farming practices, now indirectly addresses that trade-off [19,33]; however, a process-based understanding of these rela- landscape-scale processes (such as tionships is context-dependent and, therefore, highly fragmented [34,35]. Homogenization of environ- BPM) as well as the development of mental conditions typically leaves a few abundant generalists, while specialists tend to be lost [24,36]. equitable and sustainable food systems Pests and their predators react differently to the composition of the surrounding landscapes [15]. are considered [17]. fi Closing yield gaps: aims at assessing Consequently, predicting the effects of intensi cation requires careful consideration of multiple factors, differences between observed yields including landscape configuration and species characteristics [25,37,38]. This requires a broader per- and those attainable under comparable spective that includes management of landscapes. bioclimatic conditions. Differences are identified by either statistical or model- Comprehensively Measuring Agricultural Sustainability: Green Total Factor Productivity based comparisons to similar regions [4,5,62]. Critiques: (i) such comparisons The positive effect of intensifying land management on yields is well studied. The relevant range of cannot account for socioeconomic classical reductionist production functions is regarded as positively sloped in input intensification, constraints (prices for inputs and as no rational agent would purchase inputs to reduce production (Figure 1A). However, groups of outputs; access to markets, credit, and technology) and ignore impacts on operating in interlinked agricultural landscapes may find that their individual choices collec- and biodiversity; and (ii) tively reduce productivity because each operator ignores the mutualism of biodiversity and production nutritional values are considered implicit, (Figure 4.1 in [39]). Hence, the yield function in intensification may become flat or even turn negatively ‘hidden hunger’, that is, lack of sloped when BPM is considered (Figure 1B). This suggests a revision to agroeconomic models. In micronutrients, unaddressed. Ecological intensification: entails the standard models, each input is usually weighted according to its economic contribution, and an replacement of anthropogenic inputs or index of all outputs can be obtained by weighting each crop according to its share in the total eco- enhancement of crop productivity nomic value. If the output index increases faster than the input index, total factor productivity through fostering biodiversity-based (TFP) increases [40]. TFP was proposed to provide a metric for agricultural sustainability [41,42], ecological functions in agricultural practices [19]. Recent research tends to but since the output and input measures typically cover only those aspects for which markets focus on specificprocesses exist, nonmarket implications are ignored [43]. (e.g., pollination) rather than outcomes (e.g., profits) and results are presented TFP growth can be neutral or beneficial to biodiversity. For example, pest- or disease-resistant at spatiotemporal scales that are less relevant to farmers [33]. varieties can achieve the same yields with reduced use of potentially harmful chemicals. Here, Land sharing: less intense, wildlife the technology can have positive external effects (e.g., improved health or reduced chemical friendly farming at the cost of further runoff) [44]. However, new technologies are not always environmentally friendly. For example, agricultural expansion [6]. Critiques: (i) the use of the dicamba weed killer in conjunction with new varieties has led to numerous studies adopt a regional, rather than a global perspective [21]; (ii) its scale lawsuits from people who suffered collateral damage from drifting dicamba [45]. These negative dependency hampers a clear results due to the new technology would be ignored in traditional TFP approaches but would be association of landscapes to sharing or captured by a green TFP approach – also termed total productivity (TRP) [43]. Green sparing type [22]; and (iii) the sparing concept suggests to preserve TFP or TRP include negative outputs (such as pollution or biodiversity loss) and inputs based biodiversity at distant sites while on natural resources (such as groundwater or biodiversity) valued for their societal contribution compromising biodiversity in farmlands, rather than at their (often lower or zero) market value. Green TFP has been suggested as a which maintains ecosystem functions more appropriate performance measure. However, attempts at operationalization have fre- such as biocontrol or pollination (BPM concept). quently failed [39], partly due to missing indicators but also due to a lack of available modeling Land sparing: an increase of set aside concepts. Applying the BPM concept can inform estimates of green TFP [46]. land for biodiversity protection while increasing production (mostly through fi A landscape-related perspective that incorporates BPM has rarely been brought to bear in pro- intensi cation) on the remaining – managed land. duction agriculture, for at least two reasons. First, the missing indicators problem valuation of Organic farming: characterizes nonmarket inputs and outputs – is challenging because the details of the BPM relationship are management that focuses on wildlife complex and location specific. Second, relatively few individual operate at the scale where friendly farming by avoiding the use of

2 Trends in Ecology & Evolution, Month 2020, Vol. xx, No. xx Trends in Ecology & Evolution

p

TrendsTrends inin EcologyEcology & EvolutionEvolution Figure 1. Conceptual Juxtaposition of Current Agronomic Models (A) and the Biodiversity–Production Mutualism (BPM) Concept (B). In both cases, panels (A) and (B), landscape management for agriculture with all its aspects (gray box) impacts production positively (gray arrow). The relationship between input intensity levels and yields can be assumed as a saturation function [lower part of panel (A)]. The BPM concept, panel (B), specifically considers positive effects of biodiversity and ecosystem functions on yields, but also negative effects of land management on biodiversity-based ecosystem functions. In the BPM concept, however, a humpback-shaped curve can be expected due to declining yields at very high levels of input intensity.

synthetic pesticides, usually with taking a landscape-related perspective results in private gains to the owner/operator of the farm. expected lower yields or profits [18], that As shown in Figure 1, recognizing the BPM concept opens the possibility that, for intensively is, a specific farm level application of operated land, biodiversity gain may coincide with minor yield losses; or, for extensively operated agroecological principles and ecological land, substantial yield gain can occur with limited intensification if biodiversity, and hence the intensification. Critiques: (i) there is no explicit use of biodiversity, which would BPM relationship, is maintained. Indeed, gains along both dimensions may be possible [47], for in- require landscape-scale management stance, with positive effects of intercropping, as shown in a recent global-scale meta-analysis [48]. beyond farm level [23] and (ii) certification schemes vary across regions and Rethinking the BPM in Agroeconomic Models countries. Sustainable intensification (SI): is Modeling the BPM closely related to agroecological To implement and test the BPM concept, substantially revised agroecological models and their principles but suggests a multifaceted implementation in an analytical framework are required. Box 1 illustrates how the mutual feed- approach and oversees the entire food backs between biodiversity and production can be integrated into regional or global system by considering nutrition, food sovereignty, and adaptation to localities agroeconomic models. The starting point is the simulation of plant production for a given location defined by socioeconomic as well as depending on the environmental conditions and the inputs for agriculture. In order to make these environmental conditions. SI simulations dependent on ecological functions and incorporate multitrophic interactions, suitable encompasses four aspects: (i) attain scaling parameters are required that consider landscape structure and composition. higher yields, while (ii) achieving a major reduction in environmental impacts, (iii) Homogeneous spatial units providing the input parameters for yield simulations are usually de- achieve a drastic reduction in resource rived from spatially referenced intersection data on environmental conditions. For the analytical intensive foods (change diet gap), and

Trends in Ecology & Evolution, Month 2020, Vol. xx, No. xx 3 Trends in Ecology & Evolution

(iv) acknowledge a diversity of region- Box 1. Analytic Framework Deciphering the BPM specific approaches [20]. An agro-ecological and -economic framework that comprehensively accounts for the most relevant land and non-land in- Total factor productivity (TFP): has puts, including biodiversity, starts with underlying crop growth processes. Figure IA illustrates how growth of individual been considered a metric for agricultural crops depends on environmental parameters E, such as available water, nutrients, and soil fertility, but also anthropogenic sustainability [41,42]. Growth in TFP inputs, such as labor, irrigation, and fertilizer. This is illustrated by a differential equation estimating crop growth Y dynam- denotes growth in an index of all outputs ically. Crop growth could be implemented in more complex ways, that is, by distinguishing different plant organs, or in subtracted by the growth in an index of more aggregated ways, such as using regression models that do not account for intra-annual dynamics. To upscale such all inputs, influenced by changes in models to larger regions or even to the global level, such models are often repeatedly run with changing model parameters knowledge and management [40]. To depending on the location x and by using spatial maps that supply data on cropland extent, environmental conditions E, account for inputs and outputs, such as and anthropogenic input A (Figure IC). climate, soils, and biodiversity, for which no markets exist, green TFP, also Incorporating multitrophic interactions of crop growth with aboveground biodiversity in such an approach is challenging, termed total resource productivity (TRP), as biodiversity changes are driven beyond the point scale. A crop growth model that fully accounts for the BPM requires has been suggested [46]. incorporating landscape-scale properties, which are the relevant drivers of biodiversity, such as composition and config-

uration of the landscape as well as input intensity [63]. While information on input intensity LI is often available on a grid scale, such as fertilizer F, irrigation I, or labor L, the quantification of landscape composition and structure can be assessed by landscape metrics (e.g., [52]). A specific multitrophic interaction (e.g., pollination, biocontrol) defines the radius around a field in which landscape configuration, composition, as well as input intensity matters for the relevant biodiversity metrics (BD), such as presence, absence, or abundance of important species traits that provide pollination or biocontrol services. For example, insect-based pollination landscape metrics can be calculated for a radius of 250 m, 750 m, or 1 km around fields, distances of up to 3 km can be relevant for both pollination and biocontrol-providing organisms [64–66].

Biodiversity data on species’ presence or abundance in turn alters the crop growth process, either by promoting or reduc-

ing growth. Besides the well-established functions, which modify a maximum growth rate rmax given available water or nu-

trients, this maximum growth rate rmax can be adapted based on multitrophic interactions (Figure IC).

TrendsTrends inin EcologyEcology & EvolutionEvolution Figure I. Recipe for Embedding Landscape-Scale Biodiversity-Driven Ecological Functions (Multitrophic Interactions) in Global Agroeconomic Models.

4 Trends in Ecology & Evolution, Month 2020, Vol. xx, No. xx Trends in Ecology & Evolution

framework to implement BPM, these scaling parameters must additionally consider landscape structure and composition as well as negative externalities due to input intensity, which ensures the interdependence between biodiversity and yields in the yield estimates. This allows a nuanced quantitative assessment of the effects of changes in landscape parameters or input inten- sity on crop yields and biodiversity. Established model systems, like LPJml [49], InVEST (http://naturalcapitalproject.stanford.edu/software/invest), or SWAT (http://swat.tamu.edu), could serve as testbeds for implementing our analytical framework.

Worked Examples for Pollination It is estimated that in 23% of cultivated terrestrial landscapes yields are declining, most likely due to , lack of ecosystem functionality, and declining biodiversity [9]. The associa- tion of yield losses with lack of farmland biodiversity-based ecosystem functions is challenging because: (i) the use of chemicals or technical processes may compensate for the loss of ecosys- tem functions; (ii) the remaining biodiversity may provide the same ecosystem functions; and/or (iii) the negative effects of intensified cultivation methods on biodiversity and yields via BPM may occur with a long time lag [50]. A comprehensive quantitative understanding of the mutual- ism between biodiversity and production is urgently needed for both global and regional assess- ments. With few exceptions, such as pollination [51], most ecosystem functions are still poorly understood. Box 2 uses pollination as an example to illustrate how the BPM concept could be implemented for other ecosystem functions.

Knowledge Gaps for Multitrophic Interactions Data that could contribute to a better process-based understanding of ecosystem functions, quantify the BPM concept, and support implementing the analytical framework are currently being developed through data syntheses that consider landscapes, biodiversity, and agricultural management. Even though these data syntheses are garnering attention [15,52,53], agronomic indicators (management, yields) are often ignored in ecological studies; and biodiversity indica- tors are underrepresented in agronomic studies [7,25,53,54]. The most crucial knowledge gap, however, arises because ecosystem services, such as biocontrol, are complex and hence difficult to implement in larger models. Biocontrol is a crucial service relevant to all agricultural commod- ities, including staples that do not depend on animal [55]; and it applies to the control of both weeds and arthropod pests. In order to overcome these shortcomings, studies that demon- strate trait matching between pests and their natural enemies, that identify how pest densities and damage relate to landscape structure, and that investigate the relative importance of different types of pests for overall yields at global scales (including plant viruses and ) are urgently needed.

Properly implemented, models deploying the analytical framework shown in Box 1 will enable researchers to determine the circumstances in which biocontrol provides a more reliable, ro- bust, cost-effective, and ecologically sustainable form of plant protection [56]. While robust models that relate landscape structure to biocontrol are still lacking, there is evidence that a more structured landscape can provide habitat for biocontrol species [15,53,57]. These have the potential to at least partially replace commercial inputs [58] and show a positive relationship between of pollinators and biological control species, while controlling for yield performance [53].

Implications for Global Food Security A Food and Intensification Gradient Helps to Characterize Countries A nuanced picture of the BPM in global food production can be derived by characterizing all countries along a food and intensification gradient [27,28]. In regions with high income and

Trends in Ecology & Evolution, Month 2020, Vol. xx, No. xx 5 Trends in Ecology & Evolution

Box 2. Application of Yield Models That Consider the BPM for a Pollination Example Natural pollination supports production of 75% of all crops [12], and its contribution is estimated at €153 billion worldwide [67]. Pollination is critical for the production of macro- and micro-nutrients: 90% of the crops that provide vitamin C, the majority of crops that produce vitamin A, calcium, fluoride, and a large portion of folic acid producing crops are pollinated by animals [68].

Even though a complete global loss of pollination service is unlikely, using it as a thought experiment can help to provide insights into how current agroeconomic models deal with such assumptions [69]. If pollination disappears and all other inputs remain unchanged, then output in the economic model decreases by the percentage loss caused by the loss of pollination [12]. However, the resulting increase in food prices provides an incentive for intensification to mitigate the loss of production [69]. In addition, rising prices on the world market encourage production increases in other parts of the world, particularly where dependence on pollinators is less pronounced.

Given comprehensive information on how crop yields depend on the abundance of pollinating species [12,54,70], we can – in the simplest case – assume a linear re- lationship between the abundance of pollinating species and the achievable yield (Figure I). If an increase in yield is pursued via intensification, this very likely leads to a reduction in insect species richness and their abundance, which in turn reduces the pollination function [14]. It can be hypothesized that increasing agricultural intensity in landscapes with a high production of -dependent products will lead to the following pattern (Figure II): (i) yields increase with increasing intensification; (ii) after reaching a threshold (e.g., due to a decrease in pollinator abundance) yields start to decrease; (iii) the maximum level of possible yields is unknown and could also depend on spillover effects (high land-use intensity in the surrounding areas). These negative repercussions of declining biodiversity-based ecosystem function on pro- ductivity are not implemented in any global agroeconomic models used for global assessments of food security. In particular, two crucial questions remain unanswered for the pollination case: (i) are potential yields accurate? and (ii) is the nutritional value of agricultural production correctly assessed?

100 4.0

75 3.5

3.0 Pollination dependency 50 Pollination dependency Modest 10−40% Essential >90% Great 40−90% 2.5 Attained yield (%) 25 Great 40−90%

Crop yield (log(kg/ha)) Essential >90% Modest 10−40%

Small <10% 2.0 0 0 5 10 15 20 0255075100 −1 Flower−visitor density (no. 100 ) Land use intensity (%) TrendsTrends inin EcologyEcology & EvolutionEvolution TrendsTrends inin EcologyEcology & EvolutionEvolution Figure I (left). Yield Increases Differ by Crop Type of Pollinator Dependency Given Different Visitation Rates of Pollinating Species. Figure II (right). Application of Biodiversity–Production Mutualism Concept. Combining a saturating increase of yields under land-use intensification (black line) with pollination functions, which decline under intensification through increased species loss, provides an application of the BPM concept. This suggests – as a testable hypothesis – a hump-shaped relationship of pollination- dependent yields under intensification (see Figure 1 in main text).

high-input intensity (e.g., Europe and North America), food demand is unlikely to increase be- cause of negligible and advanced or completed dietary transitions. In low- income countries where input intensity is low [e.g., most of sub-Saharan Africa (SSA)], growth in population and food demand per capita is expected to be rapid. The intermediate cases (e.g., India, Indonesia) exhibit high variations in input intensity, continued population growth at a moderate and declining rate, and ongoing but incomplete dietary transitions. The vast majority of growth in global food demand over the next 30 years is expected to come from these low- income and middle-income countries.

Implications of the BPM for Low-Intensification Low-Productivity Countries Our framework suggest that in low-input areas, such as SSA, the scope for increasing production is high, as is the scope for either damaging or preserving biodiversity [10]. In these regions, large parts of the population, especially poor people, depend on agriculture for their livelihoods,

6 Trends in Ecology & Evolution, Month 2020, Vol. xx, No. xx Trends in Ecology & Evolution

allowing a dynamic agricultural sector to become an effective means of poverty reduction. The BPM concept also highlights that to ensure healthy diets, which eliminate ‘hidden hunger’ (i.e., the lack of micronutrients in food consumption), functioning ecosystems are needed to provide pollination-based commodities (Box 2). Growth in demand for food is expected to be high, with a likely concomitant growth in food production in these regions.

Consistent with our framework, grain production growth has been rapid in SSA since about 2000, especially when South Africa (high-input-intensity agriculture) and Nigeria (oil exports have slowed agricultural growth) are excluded (Figure 2). With appropriate understanding (i.e., fully considering the BPM concept), a functioning ecosystem can be a partner in the drive to improve livelihoods, reduce malnutrition, and preserve the global environment. Without this understanding, area expansion and intensification in SSA present clear threats to biodiversity. The framework suggests that accounting for BPM while seeking production increases may lead to a more favorable outcome from all perspectives.

TrendsTrends inin EcoloEcologygy & EvolutionEvolution Figure 2. Growth Rates of Cereals Production. (A) shows annualized growth rates of harvested area (dark) and production (light gray) from 2002 to 2005 and 2012 to 2015 for different world regions shaded in gray and labeled on the map. (B) displays the same annualized growth rates for selected countries along the intensification gradient: high-input intensity (red), intermediate cases (blue), and low-input intensity (green). While high-intensity countries have been able to reduce harvested area, all other regions and example countries have increased production through intensification both in terms of land use and yield per hectare. Source: Food and Agriculture Organization of the United Nations – FAOSTAT (www.fao.org/faostat/ accessed August 2018).

Trends in Ecology & Evolution, Month 2020, Vol. xx, No. xx 7 Trends in Ecology & Evolution

The Need for a Paradigm Shift in High-Input Agricultural Regions Outstanding Questions In high-input regions, less rapid growth in demand for food creates scope for reducing commer- Along a continuous gradient of land-use cial and natural resource inputs [59]. The BPM framework suggests that this would provide in- intensification, is there is a tipping point creasing production of positive externalities and reduction of negative ones. Once again, there after which key facets of biodiversity is evidence that this is happening. In Western Europe and East Asia, the growth of TFP was ac- are unavoidably lost, resulting in dimin- companied by little or no growth in agricultural output and a reduction in the overall amount of ished and potentially unstable yields? Which aspects of land-use intensifica- conventional inputs, including land, used for agriculture over the last two decades [59]. In the tion (such as energy, labor inputs, field USA, TFP growth has remained strong, while overall input use has remained flat and yield growth size, landscape configuration) deter- has slowed compared with SSA, Asia, or [60]. mine such tipping points? Knowing these major direct drivers supports identifying appropriate measures to It is important to highlight that a failure to realize adequate food production growth in low-input stop biodiversity decline in managed regions will translate into production pressure in high-input regions via trade linkages. This landscapes. observation reinforces the need for mechanisms that channel these broadly positive trends in How can high-intensity farming such a way that the production of positive externalities is increased and the production of systems be managed to support and negative externalities is minimized. In high-input environments, the concept of green TFP can pro- re-establish biodiversity? Especially vide a basis for reorienting policies, such as the subsidy system of the EU’s common agricultural landscape-scale measures, which policy (CAP), towards environmental protection and results-based incentives [61]. In low-input address farmland landscape composi- tion and configuration, are known to environments, like SSA, green TFP could help research and extension programs clarify the full enhance farmland biodiversity, but costs and benefits of alternative agricultural development pathways so that local actors can are not considered in agroeconomic make informed choices. models as well as decision making by farmers, who mostly focus on optimally managed fields and largely overlook Concluding Remarks landscape-scale consequences of The BPM concept provides the basis and a common language for different disciplines, such as their actions. agronomy, agroecology, economics, and conservation ecology, to solve a question of utmost im- What is the potential for green TFP to portance: How do we manage the terrestrial resources of our planet in such a way that we pro- quantify the extent of BPM? Which ag- duce enough healthy food without destroying our life-support system? To respond to this ricultural policy can be developed in question, it is argued here that a new metric of productivity is required – one that accounts not countries with low, intermediate, and only for all commercial inputs but also for interactions with the environment. The concept of high input intensity applying a compre- hensive measure, such as green TFP? green TFP, or TRP, is one such measure. Indeed, in 2016 the Group of Twenty (G20) commis- sioned a white paper on the subject of ‘Metrics of Sustainable Agricultural Productivity’ [39]. What are the specific key facets of This opinion article features many of the themes raised in the paper, including the importance land-use intensification for biodiversity of extending the traditional TFP measure as well as the need to link farm and landscape impacts in regions of lower land-use intensity (i.e., SSA, India, Eastern Europe)? Are in order to capture what is here called BPM. there economic or policy measures available that will reinforce these key el- The BPM concept and its modeling framework (Box 2) links agroecological principles – well ements of the landscape, thereby supporting both biodiversity and agri- known to farmers – with policy-relevant indicators, providing the quantitative base for cultural production? Can small-scale implementing green TFP and TRP. This can bridge the mismatches between scientificinterest farmers be empowered to apply such in process understanding, farmers’ interest in profitability, and decision makers’ interest in mechanisms (, permacul- measurable indicators [33]. Thus, the suggested analytical efforts can support redirection of ture, ecological intensification)? How fi can this development be supported, existing agricultural subsidies towards more economically bene cial and ecologically effective while avoiding tipping points that lead greening measures. We strongly recommend a substantial overhaul of available models and inte- to biodiversity loss and destruction of grated assessment tools. These must account for the inherent feedback between biodiversity, ecosystem functioning? ecosystem function, and resource provisioning, to be capable of identifying possible win-win op- tions in land management that maintain or even increase productivity and halt biodiversity loss (see Outstanding Questions).

Author Contributions R.S. and T.H. developed the initial idea of the conceptual framework, which was elaborated and finalized by all authors. C.A. focused on implications and perspectives for different world regions, E.A.M. and M.B. on agrobiodiversity and ecological functioning, R.S. on landscape ecology

8 Trends in Ecology & Evolution, Month 2020, Vol. xx, No. xx Trends in Ecology & Evolution

and modeling, and T.H. on global agroeconomic trade. All authors contributed equally to the development of this opinion article.

Acknowledgments The authors are grateful to Lucas Garibaldi for proving the data for Box 2 (see Figure I in Box 2), the opportunity to reanalyze these data, and the suggestions on earlier versions of this opinion article. T.H. acknowledges the support of the National Science Foundation (INFEWS CBET-1855937 and RDCEP, 386 SES-1463644) as well as the (USDA-NIFA) (IND01053G2) and Hatch (100342). M.B. acknowledges funding by the Helmholtz Research School for Ecosystem Services under Changing Land Use and Climate (ESCALATE, VH-KO-613).

References 1. Campbell, B.M. et al. (2017) Agriculture production as a major 22. Fischer, J. et al. (2014) Land sparing versus land sharing: moving driver of the Earth system exceeding planetary boundaries. forward. Conserv. Lett. 7, 149–157 Ecol. Soc. 22, 8 23. Salliou, N. et al. (2019) Governance of ecosystem services in ag- 2. Busch, J. and Ferretti-Gallon, K. (2017) What drives deforesta- roecology: when coordination is needed but difficult to achieve. tion and what stops it? A meta-analysis. Rev. Environ. Econ. Sustainability 11, 1158 Policy 11, 3–23 24. Hautier, Y. et al. (2018) Local loss and spatial homogenization of 3. Seppelt, R. et al. (2014) Synchronized peak-rate years of global plant diversity reduce ecosystem multifunctionality. Nat. Ecol. resources use. Ecol. Soc. 19, 50 Evol. 2, 50–56 4. Pradhan, P. et al. (2015) Closing yield gaps: how sustainable 25. Martin, E.A. et al. (2019) The interplay of landscape composition can we be? PLoS ONE 10, e0129487 and configuration: new pathways to manage functional biodiver- 5. Mauser, W. et al. (2015) Global production potentials sity and agroecosystem services across Europe. Ecol. Lett. 22, exceed expected future demand without the need for cropland 1083–1094 expansion. Nat. Commun. 6, 8946 26. Levers, C. et al. (2018) Archetypical patterns and trajectories of 6. Balmford, A. et al. (2015) Land for food and land for nature? land systems in Europe. Reg. Environ. Chang. 18, 715–732 Daedalus 144, 57–75 27. Václavík, T. et al. (2013) Mapping global land system archetypes. 7. Beckmann, M. et al. (2019) Conventional land-use intensification Glob. Environ. Chang. 23, 1637–1647 reduces species richness and increases production: a global 28. van Asselen, S. and Verburg, P.H. (2012) A land system repre- meta-analysis. Glob. Chang. Biol. 25, 1941–1956 sentation for global assessments and land-use modeling. Glob. 8. Newbold, T. et al. (2015) Global effects of land use on local ter- Chang. Biol. 18, 3125–3148 restrial biodiversity. Nature 520, 45–50 29. Hudson, L.N. et al. (2016) The database of the PREDICTS 9. Díaz, S. et al. (2019) Summary for Policymakers of the Global (projecting responses of ecological diversity in changing terres- Assessment Report on Biodiversity and Ecosystem Services of trial systems) project. Ecol. Evol. 7, 145–188 the Intergovernmental Science-Policy Platform on Biodiversity 30. Hallmann, C.A. et al. (2017) More than 75 percent decline over and Ecosystem Services, Intergovernmental Science-Policy 27 years in total flying insect biomass in protected areas. PLoS Platform on Biodiversity and Ecosystem Services (IPBES) ONE 12, e0185809 Secretariat 31. Seibold, S. et al. (2019) Arthropod decline in grasslands and for- 10. Tscharntke, T. et al. (2012) Global food security, biodiversity ests is associated with landscape-level drivers. Nature 574, conservation and the future of agricultural intensification. Biol. 671–674 Conserv. 151, 53–59 32. Emmerson, M. et al. (2016) How agricultural intensification af- 11. Springmann, M. et al. (2018) Options for keeping the food sys- fects biodiversity and ecosystem services. Adv. Ecol. Res. 55, tem within environmental limits. Nature 562, 519–525 43–97 12. Klein, A.-M. et al. (2007) Importance of pollinators in changing 33. Kleijn, D. et al. (2019) Ecological intensification: bridging the landscapes for world crops. Proc. Biol. Sci. 274, 303–313 gap between science and practice. Trends Ecol. Evol. 34, 13. Lechenet, M. et al. (2017) Reducing pesticide use while preserving 154–166 crop productivity and profitability on arable farms. Nat. 3, 1–6 34. Jiang, J. et al. (2018) Predicting tipping points in mutualistic 14. Kovács-Hostyánszki, A. et al. (2017) Ecological intensification to networks through dimension reduction. Proc. Natl. Acad. Sci. mitigate impacts of conventional intensive land use on pollinators U. S. A. 115, E639–E647 and pollination. Ecol. Lett. 20, 673–689 35. Wilby, A. and Thomas, M.B. (2002) Natural enemy diversity and 15. Karp, D.S. et al. (2018) Crop pests and predators exhibit incon- pest control: patterns of pest with agricultural inten- sistent responses to surrounding landscape composition. Proc. sification. Ecol. Lett. 5, 353–360 Natl. Acad. Sci. U. S. A. 115, E7863–E7870 36. Gossner, M.M. et al. (2016) Land-use intensification causes 16. Therond, O. et al. (2017) A new analytical framework of farming multitrophic homogenization of grassland communities. Nature system and agriculture model diversities. A review. Agron. 540, 266–269 Sustain. Dev. 37, 21 37. Tscharntke, T. et al. (2016) When natural habitat fails to enhance 17. HLPE (2019) Agroecological and Other Innovative Approaches for – five hypotheses. Biol. Conserv. 204, and Food Systems That Enhance Food 449–458 Security and Nutrition, Food and Agriculture Organization (FAO) 38. Perović,D.J.et al. (2018) Managing biological control services 18. Tuck, S.L. et al. (2014) Land-use intensity and the effects of or- through multi-trophic trait interactions: review and guidelines ganic farming on biodiversity: a hierarchical meta-analysis. for implementation at local and landscape scales: multi-trophic J. Appl. Ecol. 51, 746–755 traits and multi-scale filters. Biol. Rev. 93, 306–321 19. Bommarco, R. et al. (2013) Ecological intensification: harnessing 39. Fuglie, K. et al. (2016) G20 MACS White Paper: Metrics of ecosystem services for food security. Trends Ecol. Evol. 28, Sustainable Agricultural Productivity, Organisation for Economic 230–238 and Development 20. Garnett, T. et al. (2013) Sustainable intensification in agriculture: 40. Fuglie, K. (2018) R&D capital, R&D spillovers, and productivity premises and policies. Science 341, 33–34 growth in world agriculture. Appl. Econ. Perspect. Policy 40, 21. Hertel, T.W. et al. (2014) Global market integration increases like- 421–444 lihood that a future African could increase crop 41. Lynam, J.K. and Herdt, R.W. (1989) Sense and sustainability:

land use and CO2 emissions. Proc. Natl. Acad. Sci. U. S. A. 111, sustainability as an objective in international agricultural research. 13799–13804 Agric. Econ. 3, 381–398

Trends in Ecology & Evolution, Month 2020, Vol. xx, No. xx 9 Trends in Ecology & Evolution

42. Coomes, O.T. et al. (2019) Leveraging total factor productivity 56. Snyder, W.E. (2019) Give predators a complement: conserving growth for sustainable and resilient farming. Nat. Sustain. 2, natural enemy biodiversity to improve biocontrol. Biol. Control 22–28 135, 73–82 43. Gollop, F. and Swinand, G.P. (2001) Total resource produc- 57. Rusch, A. et al. (2016) Agricultural landscape simplification re- tivity. Accounting for changing environmental quality. In New duces natural pest control: a quantitative synthesis. Agric. Developments in Productivity Analysis (Hulten, C.R. et al., Ecosyst. Environ. 221, 198–204 eds), pp. 587–608, University of Chicago Press 58. Pretty, J. (2018) Intensification for redesigned and sustainable 44. Ahmed, A.U. et al. (2019) Impacts of Bt Brinjal (Eggplant) agricultural systems. Science 362, eaav0294 Technology in Bangladesh, International Food Policy Research 59. Heisey, P.W. and Fuglie, K.O. (2018) Agricultural Research Institute Investment and Policy Reform in High-Income Countries, 45. Nickel, R. and Polansek, T. (2018) Battle of the : Monsanto United States Department of Agriculture faces a fight for soy market, Reuters Business News 24 January 60. Grassini, P. et al. (2013) Distinguishing between yield advances 2018. https://www.reuters.com/article/us-usa-pesticides-soy- and yield plateaus in historical crop production trends. Nat. beans-insight/battle-of-the-beans-monsanto-faces-a-fight-for- Commun. 4, 2918 soy-market-idUSKBN1FD0G2 61. Pe’er, G. et al. (2017) Adding some green to the greening: im- 46. Ehui, S.K. and Spencer, D.S.C. (1993) Measuring the sus- proving the EU’s ecological focus areas for biodiversity and tainability and economic viability of tropical farming sys- farmers: evaluation of EU’s ecological focus areas. Conserv. tems: a model from sub-Saharan Africa. Agric. Econ. 9, Lett. 10, 517–530 279–296 62. Mueller, N.D. et al. (2012) Closing yield gaps through nutrient 47. Kremen, C. and Merenlender, A.M. (2018) Landscapes that and water management. Nature 490, 254–257 work for biodiversity and people. Science 362, eaau6020 63. Graham, L.J. et al. (2019) Incorporating fine-scale environmental 48. Li, C. et al. (2020) Syndromes of production in intercropping im- heterogeneity into broad-extent models. Methods Ecol. Evol. 10, pact yield gains. Nat. Plants 6, 653–660 767–778 49. Schaphoff, S. et al. (2018) LPJmL4 – a dynamic global vegeta- 64. Steffan-Dewenter, I. et al. (2001) Pollination, set and seed tion model with managed land – part 1: model description. on a landscape scale. Proc. R. Soc. B Biol. Sci. 268, Geosci. Model Dev. 11, 1343–1375 1685–1690 50. Kuussaari, M. et al. (2009) debt: a challenge for biodi- 65. Steffan-Dewenter, I. et al. (2002) Scale-dependent effects of land- versity conservation. Trends Ecol. Evol. 24, 564–571 scape context on three pollinator guilds. Ecology 83, 1421–1432 51. Priess, J.A. et al. (2007) Linking scenarios to polli- 66. Hass, A.L. et al. (2018) Landscape configurational heterogeneity nation services and economic returns in coffee agroforestry sys- by small-scale agriculture, not , maintains pollina- tems. Ecol. Appl. 17, 407–417 tors and plant reproduction in Western Europe. Proc. R. Soc. 52. Dominik, C. et al. (2018) Landscape composition, configuration, B Biol. Sci. 285, 20172242 and trophic interactions shape arthropod communities in rice 67. Lautenbach, S. et al. (2012) Spatial and temporal trends of agroecosystems. J. Appl. Ecol. 55, 2461–2472 global pollination benefit. PLoS ONE 7, e35954 53. Dainese, M. et al. (2019) A global synthesis reveals 68. Eilers, E.J. et al. (2011) Contribution of pollinator-mediated biodiversity-mediated benefits for crop production. Sci. Adv. crops to nutrients in the human food supply. PLoS ONE 6, 5, eaax0121 e21363 54. Garibaldi, L.A. et al. (2016) Mutually beneficial pollinator diversity 69. Bauer, D.M. and Sue Wing, I. (2016) The macroeconomic cost of and crop yield outcomes in small and large farms. Science 351, catastrophic pollinator declines. Ecol. Econ. 126, 1–13 388–391 70. Garibaldi, L.A. et al. (2015) Trait matching of flower visitors and 55. Losey, J.E. and Vaughan, M. (2006) The economic value of eco- crops predicts fruit set better than trait diversity. J. Appl. Ecol. logical services provided by . BioScience 56, 311 52, 1436–1444

10 Trends in Ecology & Evolution, Month 2020, Vol. xx, No. xx